The correct way to bond shields - Resolution

technology
The correct way to bond
cable shields
Recent coverage of the topic in the pages of this magazine have begged explanation of the whole double-ended shield bonding issue.
The knowledge base and the substantiating proof exists, according to KEITH ARMSTRONG and TONY WALDRON.
WHEN WE WERE LEARNING how to do
pro audio decades ago, we learnt that
experienced engineers minimised the noise
in a mixing console by spending a great deal of time
‘fiddling about’ with the terminations of its cable
shields and its internal bonding.
Despite the fact that the console’s electronics
modules worked satisfactorily on the bench, they
behaved quite differently when connected together in
a console frame – and differently again when finally
installed on the customer’s site. The main problem was
interference known as ‘hum’ or ‘buzz’ and the purpose
of the lengthy fiddling about was to ‘eliminate the
hum loops’ (sometimes called ground loops).
In recent years we have discovered this expensive
‘fiddling about’ was all unnecessary. Because of a
design decision made many decades ago, possibly
even for no very good reason, and repeated as if it
were gospel ever since, single-point grounding
techniques and single-ended cable shield bonding
became the standard for equipment design. As a direct
result of this poor decision pro audio equipment and
systems were condemned to suffer hums and buzzes
requiring costly devices with high common-mode
rejection ratio (CMRR) and/or time-consuming
‘fiddling about’ to solve.
It took the introduction of digital technology and the
EMC directive with its EMC standards EN 55103-1 and
-2 to force us to abandon our decades-old practices.
Now we use mesh-bonding techniques and bond the
shields of our cables to chassis at both ends – a process
of discovery described [1] by Tony Waldron.
new and legacy pro audio installations have already
benefited greatly from the use of our new techniques.
Equipment should be designed so as to bond the
shields of every cable (digital or analogue) to its local
chassis or frame – ideally to the surface of its Faraday
cage enclosure shield, if it has one. This is not a recent
revelation – the Audio Engineering Society devoted a
full issue of its Journal to it in 1995 [5], and John
Watkinson described its benefits in Studio Sound in
1995 [2], in 1996 [3], and most recently in the
July/August 2002 issue of Resolution [4].
Next, both ends of every cable shield should be
bonded directly to their local chassis or Faraday cage
as specified by IEC 61000-5-2 [6]. Of course, this
causes ground loop currents to flow in the cable
shields, but we have recently shown [7] that this
cannot cause significant audio noise – even in the very
highest quality systems – even with shield currents
large enough to heat the cables.
Our article [7] in the EMC Compliance Journal
describes tests on a variety of balanced audio cables
nearly 30m long to determine the effects of powerfrequency
shield currents on audio
noise. Several types of balanced cables
were tested, including an extremely poor
quality cable with untwisted signal
conductors and a capacitive imbalance
exceeding 20% – representing the worst
legacy cabling that can be expected.
No properly managed pro audio system
has ground potential differences large
enough to drive damaging currents in
cable shields, although they can occur in
heavy industry and during ground-faults.
In any case, the solution is simply to run
a parallel earth conductor (PEC) between
the offending items of equipment, as
described in IEC 61000-2-5 [6].
Figure 2 shows the results from one of the tests
described in [7], simulating a floating 75 ohm signal
source. The CMRR performance of an audio system
when using four different types of cables (B, G, E, F) is
shown, cable G being the extremely poor quality cable.
The noise performance of the cables with 5 amps of
shield current is much the same as when their shields
are one-end bonded. Adding the 6mm2 PEC reduced
the noise by about 10dB, so it can be beneficial to
employ a PEC even when shield currents are low.
Similar results were obtained for a variety of other
signal sources.
Real pro audio systems have large numbers of
cables, and when their shields are all bonded at both
ends their combined impedance is very low. Since the
source impedance of ground potential differences is
finite, the more shields are bonded at both ends and/or
the more PECs that are employed, the lower is the
resulting ground noise potential. This is shown by
Figure 3 for a floating 75 ohm signal source, taken
from a series of such graphs in [7].
Tony Waldron and Cadac R-Type
We were surprised and gratified to discover that our
new design techniques eliminated all the usual
‘fiddling about’ with cable shields and grounding –
saving substantial amounts of time and money. They
simply reduce the ground potential differences
between items of equipment – which are the cause of
hum and buzz problems.
We were also very pleased to find that our new
techniques considerably improved audio performance;
avoided the need for expensive high-CMRR devices;
and improved equipment reliability and safety. Many
Figure 2. Cable CMRRs for a floating 75 ohm source
Figure 3. System CMRR versus ground
system impedance
Figure 3 shows the overall system CMRR
for four different configurations of the
same four cable types as Figure 1. It
shows that in real life, bonding cable
shields to chassis at both ends usually
gives much higher system CMRR (lower
system noise) than the traditional
technique of bonding cable shields at
only one end.
Note that although the amplifiers used
in [7] had a CMRR of well under 100dB,
the overall system CMRR can exceed
140dB. This is one reason why pro audio
systems achieve better quality sound
when bonding shields to chassis at both ends, without
68 resolution September 2002

technology
the need for costly high-CMRR devices such as special
transformers.
Bonding cable shields to chassis with ‘pigtails’ (a
length of wire and/or a connector pin), as XLR and
many other standard audio connectors often force us
to do, is sufficient for eliminating hum and buzz
interference caused by mains frequencies, but useless
at preventing radio-frequency (RF) interference.
Controlling the emissions of, and susceptibility to,
interference is known as EMC. Most readers will know
that all pro audio, video, and entertainment lighting
equipment and systems supplied in Europe must by
law meet the EU’s EMC Directive by applying the
harmonised standards EN 55103-1 and EN 55103-2,
affixing the CE mark and having a signed Declaration
of Conformity. (An alternative is known as the
Technical Construction File route.)
Designing equipment to employ RF shield
terminations at both ends of typical shielded audio
cables helps achieve EMC compliance at the lowest
cost, by considerably reducing the size and cost of the
filters that would otherwise be needed at every input
and output. Happily, some audio connector
manufacturers can now supply products that use
reasonable RF shield bonding techniques.
Some people have been recommending capacitive
shield bonding at one cable end to help achieve good
EMC, but this would still require ‘fiddling
about’ and the use of expensive high-
CMRR devices for the highest quality.
For good EMC performance to 1000MHz
as required by EN 55103-1 and -2 the
special capacitive connectors required
would be costly, and will often need to
be fitted at both ends of the cable to
permit the traditional time-consuming
‘fiddling about’.
Poor equipment design is the real
cause of ‘hum loop’ noise. Shield
currents have got such a bad reputation
for causing noise because most legacy
equipment was designed using very
poor shield bonding practices, such as
the method shown in Figure 4.
Figure 4. It is poor equipment design that causes
‘ground loop’ noise
Ground loop currents in cable shields are injected into
the finite impedances of a sensitive audio circuit’s 0V
system, causing noise voltages to arise in the 0V
references of the unbalanced audio circuits and
making their signals noisy. Very careful use of singlepoint
grounding can go some way to reducing this
problem, but the stray magnetic coupling between
wires and PCB traces means that shield currents will
always cause unacceptable hums and buzzes.
Another traditional but very poor technique is to
use a ‘chassis’ trace (or wire) to connect all the shield
pins together and (eventually) connect them to the
chassis. These ‘chassis ground’ traces are isolated from
0V but magnetically couple with any nearby signal or
0V traces or wires, giving rise to noise voltages.
All these noise problems would be eliminated if
cable shields were connected directly to the
chassis/frame/enclosure at the point where their cable
enters the equipment (as recommended by [2], [3],
[4], [5] and [6]) and this is also best for EMC. A
surprising proportion of new equipment still uses longdiscredited
shield bonding methods.
Although grounding system resistances are usually
less than 2 ohms, their inductance makes their
impedance to lightning and other types of voltage
surge much higher.
Reference [8] shows that the effect is to apply most
of the surge voltage to the electronic devices at cable
ends where shields are not bonded to chassis, and this
Figure 5. Surge overvoltages caused at
signal inputs and outputs
is very bad for reliability (Figure 5).
Bonding cable shields at both ends and
using PECs where appropriate
considerably reduces the inductance of the
grounding system, reducing the surge
voltages that develop as well as helping to
protect input and output devices from
overvoltages.
The ends of cables that have
unterminated shields have occasionally
been seen to flash-over during
thunderstorms, so there are clearly
significant shock and fire hazards
associated with bonding cable shields at one end.
Finally, it is a common practice in the pro audio industry
to disconnect the protective ground conductors from
items of equipment to help ‘cure’ hum loops quickly and
at low cost. This is a very bad practice indeed, and is
illegal under health and safety legislation in many
countries. When cable shields are bonded at both ends
this traditional (but unsafe) remedy is not required.
In conclusion, bonding the shields of balanced
cables at only one end only achieves good noise
performance for complex installations after a great deal
of costly ‘fiddling about’, and also wastes a low-cost
EMC resource.
Concerns about ‘hum loop’ noise when bonding both
ends of the shields of balanced audio cables are without
basis for correctly designed equipment. Correct equipment
design need cost little more and helps achieve EMC
compliance at lowest cost by using the EMC performance
of existing cable shields. As the number of cable shields
bonded at both ends in an installation increases, and/or as
more PECs are used, the noisy ground potential
differences in an installation fall and so do the resulting
CM noise voltages. The large number of shielded cables
available in a typical pro audio installation make it easy to
reduce ground noise potentials dramatically –
substantially increasing the system’s overall CMRR.
Even where EMC compliance is not a necessity,
bonding cable shields at both ends saves large
amounts of time and cost by totally eliminating any
need to ‘fiddle about’ with hum loops. It also improves
sound system quality, reliability, and safety. ■
Contacts
Tony Waldron is technical manager for Cadac and former
sound designer at The Royal Danish Theatre, Copenhagen
and head of sound at The National Theatre, London.
Tel: +44 1582 404202
E-mail: tony@cadac-sound.com
Website: www.cadac-sound.com
Keith Armstrong specialises in EMC and safety
compliance for Cherry Clough Consultants and previously
designed high-performance analogue circuits, A-D and D-A
convertors for Neve
mixing desks.
Tel: +44 1457 871605
E-mail: keith.armstrong@cherryclough.com
Website: www.cherryclough.com
1 Designing for interference-free audio system
components, Tony Waldron, EMC Compliance
Journal, July 2002, p22 (www.complianceclub.com)
2 EMC: who’s really afraid now? John Watkinson,
Studio Sound, October 1995, p13
3 Money for old cabling, John Watkinson, Studio
Sound, September 1996, p89
4 The cable snake, John Watkinson, Resolution,
July/August 2002, p57
5 Journal of the AES: Volume 43, Number 6,
1995, Special Excerpt – Shields and Grounds.
6 IEC 61000-5-2:1997 Electromagnetic
compatibility (EMC) – Part 5: Installation and
mitigation guidelines – Section 2: Grounding
and cabling
7 Bonding cable shields at both ends to reduce
noise, Tony Waldron and Keith Armstrong, EMC
Compliance Journal, May 2002, p14
8 Bonding the screens of cables at both
ends, Keith Armstrong, ERA Conference
‘Earthing 2000’ 21-22 June 2000, Solihull
(www.eratechnology.co.uk)
September 2002
resolution
69